Metadata-Version: 2.4
Name: rifft-sdk
Version: 0.1.1
Summary: Cross-framework debugger for multi-agent AI systems
Author-email: rifft-dev <hello@rifft.dev>
License: MIT
Project-URL: Homepage, https://github.com/rifft-dev/rifft
Project-URL: Repository, https://github.com/rifft-dev/rifft
Project-URL: Issues, https://github.com/rifft-dev/rifft/issues
Keywords: ai,agents,debugging,observability,opentelemetry
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Requires-Python: >=3.9
Description-Content-Type: text/markdown
Requires-Dist: opentelemetry-sdk>=1.30.0
Requires-Dist: opentelemetry-exporter-otlp>=1.30.0

# rifft-sdk

Python SDK for [Rifft](https://github.com/rifft-dev/rifft), a cross-framework debugger for multi-agent AI systems.

`rifft-sdk` lets you instrument Python agents, spans, handoffs, tool calls, and decisions, then view the resulting trace in Rifft with communication graphs, timelines, replay, and MAST failure classification.

## Install

```bash
pip install rifft-sdk
```

## Quickstart

```python
import rifft

rifft.init(service_name="my-agent-app")

@rifft.trace(agent_id="researcher", framework="custom")
def run_research():
    with rifft.span("tool.call"):
        return {"summary": "Agent trace debugging helps teams inspect handoffs."}

run_research()
```

By default, the SDK exports OTLP traces to the local Rifft collector.

## What You Get

- trace capture for Python multi-agent workflows
- agent and tool spans with timing and cost metadata
- communication edges and handoff visibility
- compatibility with Rifft adapters such as CrewAI, AutoGen, and MCP

## Links

- Repository: https://github.com/rifft-dev/rifft
- Issues: https://github.com/rifft-dev/rifft/issues
